Populations usually persist despite environmental variations. Experimental analysis of responses to stress must include distinction between potential stresses (environmental perturbations that might not cause stress) and actual stress (phenomena that cause a response by the population). This is made difficult by large temporal fluctuations in abundances of many organisms. Monitoring can measure this variability but is insufficient to predict the potential impact of most stresses. Experimental analyses of stresses are also made difficult by differences among populations in their inertia (lack of response to perturbation), resilience (magnitude of stresses from which a population can recover) and stability (rate of recovery following a stress). These attributes of populations cause a range of responses to intermittent, temporary and acute (or ‘pulse’) stresses and to long-term, chronic (‘press’) disturbances. The timing, magnitude and order of stresses can cause different responses by populations. Synergisms between simultaneous or successive stresses can also have unpredictable effects on populations and cause complexity in interpretations of patterns of competition and predation. Experimental manipulations are needed to understand the likely effect of environmental disturbances on populations. The appropriate experiments are those designed to measure the effects of different types, magnitudes and frequencies of simulated stresses. These will be more revealing than the more common experimental analyses used to determine why and how observed changes in abundances of populations are caused by existing stresses.